What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 130 Stack Exchange communities.

I was bored this afternoon and decided to do something with the stack overflow data dump. I defined a "Jon Skeet" number, similar to the Erdos Number.

You can check your Jon Skeet Number here: http://sharpenyourteeth.net/stackoverflow/index.php

The Erdos Number is based on authorship of papers. To calculate a Jon Skeet number I define the "authorship" of answers to a particular question. First, only questions that are not closed and not community wiki are considered. A question must also have an answer of at least 5 upvotes (doesn't matter if there is an accepted answer or not). The "authors" that answered that question are determined as follows. Say the top answer has X points. All users with an answer in the range [ceil(0.75*X), X] are considered co-authors to the answer of that question.

To actually calculate the Jon Skeet number I made a graph of authors (there is an edge between every pair of co-authors) and did a breadth-first search starting with Jon Skeet.

Waste of time? Yeah, but I haven't written any code in a while so it was fun. Maybe someone else will like the idea.

share|improve this question

closed as not constructive by Lance Roberts, Tim Post Aug 7 '11 at 18:38

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.If this question can be reworded to fit the rules in the help center, please edit the question.

2  
Congrats the creator of the 'statistics' tag, you just became a Taxonomist. –  Ether Jan 11 '10 at 0:16
    
I got a 144. :) –  Tyler Carter Jan 11 '10 at 0:20
2  
Excellent work. I think it would be interesting to see the graph. –  Bill the Lizard Jan 11 '10 at 3:27
2  
Well done! But now I want to know my Bacon-Skeet number. –  zombat Jan 11 '10 at 5:01
    
@Bill the Lizard: I thought so too, but the original graph has around 6k nodes and 20k edges. I tried limiting it to users with a number less than 30 and only showing the graph after BFS (removing unnecessary edges). This is a more manageable ~800 edges. Here is what that looks like sharpenyourteeth.net/stackoverflow/graph.pdf (graph of user IDs). If I come up with a more interesting visualization I'll post it. –  Neal Jan 11 '10 at 5:19
    
[ceil(0.75*X, X]? Did you miss a closing bracket? –  Sam152 Jan 11 '10 at 11:59
1  
I could have had an Erdos number of 4 if I'd ever published a paper in college. I believe that statement awards me 0.0025 geek cred points. –  mmyers Jan 11 '10 at 15:33
2  
Pretend I don't have a clue of what are you talking about and pretend I didn't understand anything from your explanation. In simple and more mundane words: What is this number???? –  OscarRyz Jan 11 '10 at 16:19
    
Infinity :( I've never felt so badly connected. –  nbolton Jan 12 '10 at 8:12
    
Infinity :( :( :( –  William David Edwards Aug 1 at 20:10

16 Answers 16

I see only two questions in your post. The answers are: 0; and yes, but I approve :)

share|improve this answer
19  
"I'm Jon Skeet, and I approve this message." +1 –  nbolton Jan 12 '10 at 8:11

All I get is “your number is Infinity”.

Your number is Infinity

share|improve this answer
1  
That means there is no path from Jon Skeet to you on the graph. You haven't co-authored an answer with someone who co-authored, with someone who co-authored, etc, with Jon Skeet. –  Neal Jan 11 '10 at 13:15
7  
It probably means that you have absolutely nothing in common with Jon Skeet. But don't worry, you are not alone! –  awe Jan 11 '10 at 13:21
3  
another guy with "Your number's Infinity" –  Sathya Jan 11 '10 at 14:12
    
So, the obvious follow-up question: What are the statistical properties (number, "average" size, etc.) of cliques (the mathematical term) of co-author closures other than the big-ass Jon Skeet clique? –  Tom Hawtin - tackline Jan 11 '10 at 15:25
2  
"Your number's Infinity" Perhaps the more obvious question: Which infinity? –  Tom Hawtin - tackline Jan 11 '10 at 23:58
1  
Maybe you should make it write You don't have a path, or something like that, rather than Your number is Infinity? :p –  Svish Jan 12 '10 at 12:02
1  
Another one for Infinity... –  Heather Jan 12 '10 at 22:56
    
Add me too to infinity list. :) –  EAGER_STUDENT Jul 20 '13 at 13:53

I get a one, but I think the real winners are the block of 42. Or the people with, you know, lives...

share|improve this answer

I get a 1... I think. But my username is not unique, so it might be another Richard.

Perhaps the user id should be used, there are unique.

share|improve this answer
    
Updated the page to include an input for user id. You still have a 1 :) –  Neal Jan 11 '10 at 13:38

Jon Skeet may be the "Erdos" of Stackoverflow, ie, the most prolific author of answers.

However, some research should be done to find out if he's the most prolific co-author. I suspect that once he has answered a question few others with his knowledge and experience choose to add their own answer.

We see there are 363 questions he co-authored an answer on. Is he the person with the highest number of co-authored answers?

At any rate, 12, and yes - but entertaining.

share|improve this answer
1  
He is indeed the person with the highest number of co-authored answers. You can see the complete list here: sharpenyourteeth.net/stackoverflow/num_coauthors.txt First column is user id, second is how many co-authors they have (i.e. edges in the graph from that user). –  Neal Jan 11 '10 at 17:36
    
Cool, thanks for the update! –  Adam Davis Jan 11 '10 at 19:40
    
I've co-authored only 40 answered questions with others, 1/9th of what Jon has, putting me at #42 - a very nice round number. –  Adam Davis Jan 11 '10 at 19:50

Neat idea - apparently my "Jon Skeet number" is 255, which is understandably high since there's basically no overlap in the types of questions Jon and I answer. Oh, and you might want to provide a lookup based on the user ID number, as usernames aren't unique.

share|improve this answer

How many valid questions (that meet your scoring criteria) are in each tier (e.g. how many questions directly involve Jon Skeet, then how many questions are one level away from that, etc etc)?

Do you consider users X and Y to have a relationship (i.e. have an edge) if one wrote the question and the other wrote an answer? (e.g. would users have an edge with Jon Skeet if they answered a question he asked?)

  • 324
  • absolutely, but perfectly appropriate for meta :)
share|improve this answer
    
I can't answer your first question (without additional processing/coding). The final graph constructed is based on co-authors, not which questions they've answered. I can tell you how many people are at each Jon Skeet level though. See: sharpenyourteeth.net/stackoverflow/dist.txt For your second question, no. There is only an edge between X and Y if they were "co-authors" (for my definition of co-authors above) for the answer to a particular question. –  Neal Jan 11 '10 at 0:35
    
That's a really interesting distribution, and not at all what I expect. There are bursts every dozen questions or so where it appears that suddenly the circle gets bigger, where lots of answers (and therefore users) suddenly become eligible to be added to the graph. What I find odd is all the other users who post heavily in my predominant tag ([perl]) have much lower numbers (e.g brian d foy and Michael Carman at 183, Sinan Unur at 94), rather than being clustered closer to my region. I suppose the question at which I become eligible is not the type of question I would have thought. –  Ether Jan 11 '10 at 5:15

Mine's 1.

Now what does it mean?, I don't have a clue. Although I have read your description a number of times, and I have ( started ) to read the Erdos article in the Wikipedia ( I stopped when I asked my self, am I procrastinating now? Should I be coding instead of reading this "Erdos" number description? ) I still don't know what is it.

So, in a simple-wiki fashion ( wait minute, why don't I query that .... let's see... )

Here it is: Erdos Number @ Simple Wikipedia ( and I thought simple wikipedia that was stupid )

So, this means that I actually collaborate with Jon Skeet? ( Would it be on the early StackOverflow stages where I reformatted all his Java answers because he used C# braces style instead of Java's?, I stop doing that when Jon Skeet facts came out and I was afraid of getting my Mac exploded by Jon Skeet mind waves!!! )

share|improve this answer
    
Here is the question you co-authored with Jon Skeet: stackoverflow.com/questions/401415 The question just barely meets the requirements (5 upvotes for best answer, which was Jon's). Your answer has at least .75 times as many upvotes as the highest, so you're a co-author with Jon, leading to a Jon Skeet number of 1. –  Neal Jan 11 '10 at 17:33
    
So, co-authoring is not necessarily "editing" the same answer , but answer the same question. Mhh interesting. I should have dozens of this question, for Jon's is very active in Java –  OscarRyz Jan 11 '10 at 18:18

I get to 9 which means my contributions overlap with Mr Skeet's a lot less than Kyle Cronin's do. Which is moderately unsurprising to me - I have the C# questions ignored (since I don't use C# at all, and have approximately zero interest in learning it as a language).

Hmmm...well, I'm not sure, now, what 9 means. The Erdős Number is the distance between two (mathematical) authors; people with the value one have written a paper with Paul Erdős; people with a value 2 have do not qualify for the value 1 but have written a paper with someone who does qualify for value 1, and so on.

So, I guess that means there are 8 intermediaries between me and Mr Skeet for questions.

Answering the questions in the original post:

  • 9
  • Yes
share|improve this answer
    
Yeah - see my rewrite. I'm intrigued that you manage to keep so far at arm's length from Jon. –  Jonathan Leffler Jan 11 '10 at 0:13

My number ended up being 60, which I thought at first was surprisingly low. Then I realized that the related tag which most often appears with matlab is java (39 questions), which is obviously one of the tags within which Jon displays juggernaut-like dominance. ;)

share|improve this answer

I got a 1. Weird.

share|improve this answer
    
Here's one of the answers you "co-authored" with Jon Skeet, giving you a 1: stackoverflow.com/questions/479883 –  Neal Jan 11 '10 at 2:55
    
Oh, so all it takes is one question? My bad. Not weird at all. –  BFree Jan 11 '10 at 3:42

I got 758. I'm unclear whether that's good or bad...

share|improve this answer

I got 364. I honestly expected it to be much higher since I never even knew who Jon Skeet was until recently and don't do much with .Net anymore.

share|improve this answer

Mine was 315, which I found very interesting, and with Stephen Darlington's 758 this is even more interesting because Erdos numbers tend to be very low: I certainly didn't publish with a wide range of co-authors, yet my Erdos number is only 5. I would guess one reason is that there aren't that many "co-authored" answers on Stack Overflow. In scientific publishing, very few people publish all on their own, and students get low Erdos numbers through their advisors. But here, I seem to have only 3 co-authored answers out of 84 total answers, so the graph gets deep rather than wide.

It would be interesting to see some statistics on the graph, like sizes of connected components, and the diameter of at least Jon Skeet's component (which I assume is the largest one?). Also, adjusting the concept of co-authorship could be interesting: for instance, what happens if every answer with a positive score is counted for co-authorship credit?

share|improve this answer
    
Also kind of funny to have such a big number, considering that my top tag is java, but I guess none of those qualify for co-authors or something. –  jk. Jan 11 '10 at 19:01
    
Here are a few stats on the graph: 8089 nodes and 13353 (undirected) edges. The are a total of 539 (!) connected components. The one containing Jon Skeet has the majority of nodes though: 6908. The remaining components have on average 2 nodes in them. My little laptop couldn't compute the diameter with nearly 7k nodes. If I have time I'll look into more efficient approaches to compute it. Lastly, I did try different levels of co-authorship. At some point I tried any non-negative answer, which took too long for the program to run so I didn't bother (too long being > 5 minutes in Python...) –  Neal Jan 11 '10 at 22:52

I got a 30. I think it's because if I see a question that Jon answered, I don't bother =P

share|improve this answer

535... so I suppose that isn't too "good". However, if you make one that count how many of my questions that has been answered by Jon Skeet, and how many questions in comments I have had answered by Jon Skeet, I think I would get a pretty high score :p

I sometimes feel like Jon Skeet and Marc Gravell are stalking me to answer my questions... which isn't too bad actually, since answer fast and rarely wrong :p So, thank you both and all the others that answers my numerous questions! I've been learning a lot thanks to StackOverflow this last year :D

share|improve this answer

Not the answer you're looking for? Browse other questions tagged .